在大查询中按月和年分组的日期频率中填充零

Question

我有一张表，上面有hiredate (Date)、First Name (String) 和 Sur Name (string)，如下所示：

hireDate    First Name      Surname
13-oct-14   Cintia Roxana   Padilla Julca
28-oct-14   Conor           McAteer
28-oct-14   Paolo           Mesia Macher
28-oct-14   William Anthony Whelan
15-nov-14   Peter Michael   Coates
13-feb-15   Natalie         Conche
15-mar-15   Beatriz         Vargas Huanca
01-may-15   Walter          Calle Chenccnes
04-may-15   Sarah Louise    Price

我在另一列中查看了hire_dates(DATE) 的频率和累积频率，如下所示：

Row hireDate    Count       Cumulative
1   13/10/2014  1           1
2   28/10/2014  3           4
3   15/11/2014  1           5
4   13/02/2015  1           6
5   15/03/2015  1           7
6   09/04/2015  1           8
7   15/04/2015  1           9
8   01/05/2015  1           10

查询是这样的：

WITH
Data AS (
 SELECT
 hireDate,
 COUNT(1) AS Count
 FROM
 `human-resources-221122.human_resources.employees_view`
 WHERE
 status <> "cancelled"
 GROUP BY
 1 )

SELECT
hireDate,
Count,
SUM(Count) OVER (ORDER BY hireDate ASC ROWS BETWEEN UNBOUNDED PRECEDING 
AND CURRENT ROW) AS Cumulative
FROM
Data
ORDER BY
hireDate ASC

但我需要在那些没有计数的地方按月份和年份查看数字，例如：

Hire_Month  Hire_Year   Count   Cumulative
October     2014        4       4
November    2014        1       5
December    2014        0       5
January     2015        0       5
February    2015        1       6
March       2015        1       7
April       2015        2       9
May         2015        1       10

提前致谢。

Answer 1

注意使用GENERATE_DATE_ARRAY 和RIGHT JOIN 以获得所需的结果：

WITH data AS (
  SELECT * 
  FROM UNNEST ([
    STRUCT(DATE("2014-12-03") AS d, 4 AS a)
    , STRUCT("2015-01-05", 7)
    , STRUCT("2015-03-05", 1)
  ])
), all_months AS (
   SELECT month
   FROM UNNEST(GENERATE_DATE_ARRAY(
     (SELECT DATE_TRUNC(MIN(d), MONTH) FROM data)
     , (SELECT MAX(d) FROM data)
     , INTERVAL 1 MONTH)
   ) AS month
)


SELECT month, IFNULL(SUM(a),0) a, SUM(SUM(a)) OVER(ORDER BY month) a_cum
FROM data 
RIGHT JOIN all_months
ON DATE_TRUNC(d, MONTH)=month
GROUP BY month
ORDER BY month

现在，如果我们只是在计数，您可以使用 LEFT/RIGHT JOIN 将在空月份有空值这一事实。这就是查询如何适应任意表（此处为维基百科）：

WITH data AS (
  SELECT *, DATE(datehour) d
  FROM `fh-bigquery.wikipedia_v3.pageviews_2018` 
  WHERE wiki='pt'
    AND (datehour BETWEEN '2018-09-30' AND '2018-09-30'
      OR datehour BETWEEN '2018-12-01' AND '2018-12-02'
    )
    AND title LIKE 'Calif%'
), all_months AS (
   SELECT month
   FROM UNNEST(GENERATE_DATE_ARRAY(
     (SELECT DATE_TRUNC(MIN(d), MONTH) FROM data)
     , (SELECT MAX(d) FROM data)
     , INTERVAL 1 MONTH)
   ) AS month
)


SELECT month, COUNT(d) c, SUM(COUNT(d)) OVER(ORDER BY month) a_cum
FROM data 
RIGHT JOIN all_months
ON DATE_TRUNC(d, MONTH)=month
GROUP BY month
ORDER BY month